Discrimination between Holstein-derived milk and pure Jersey dairy products via analysis of the MC1R gene

Abstract

The majority of dairy cows in Japan are Holstein cows. Milk items containing only milk from Jersey cows are sold at a premium price. Therefore, it is important to keep Holstein milk from contaminating the milk labeled as pure Jersey milk to maintain brand value. Here, genetic testing was conducted on packaged milk items. Polymorphisms of the melanocortin-1 receptor (MC1R) gene, which controls variations in coat color, were assayed in addition to the frequencies of MC1R alleles in Holstein and Jersey cows in Japan. In Holstein cows, the frequency of the E^D allele was 0.876, but this allele is absent in Jersey cows. Thus, detection of the E^D allele clearly indicates the presence of Holstein DNA. This DNA-based test was also useful in testing sterilized milk products. The results of the DNA-based test show that pure Jersey milk can be easily distinguished from Holstein milk.

Introduction

DNA analysis is widely used to trace meat and dairy products to the source breed or even to the individual animal (Kijas et al., 1998; Dalvit et al., 2007; Sasazaki et al., 2007; Sasazaki et al., 2011). The European Union has enacted laws to regulate the labeling of regional foods; for example, a manufacturer cannot claim that the milk of a Holstein cow is the source of Beaufort, Abondance, or Reblochon cheeses, which are produced in highlands of the French Alps. Maudet and Taberlet (2002) successfully extracted DNA from milk used to produce cheese and demonstrated that the integration of Holstein milk is detectable by DNA analysis.

In Japan, the vast majority (∼98.5%) of dairy cattle are Holstein cows. Although Jersey is the second most popular breed, these cows comprise < 1% of the entire Japanese dairy cattle population. According to data collected in June 2020, there are 1 361 177 Holstein cows and 11 477 Jersey cows in Japan (National Livestock Breeding Center)ⁱ⁾. Jersey mono-labeled milk is considered a premium product, and its retail price is 1.5–2 times higher than that of breed-unspecified milk in Japan. The mislabeling of Jersey milk has not been reported so far, but price differences between breeds may potentially motivate food fraud. Therefore, detection of Holstein milk labeled as pure Jersey milk is important to maintain brand value. However, to the best of our knowledge, there are no reports specifying DNA-based identification for the breed traceability of milk in Japan.

Coat color is an important characteristic to identify the breed of livestock. Many breeds of livestock have typical coat colors; thus, DNA polymorphisms of pigmentation genes are attractive candidates as markers for breed identification (Kijas et al., 1998; Russo et al., 2007; Sasazaki et al., 2007). The unique black- and white-spotted coat markings are characteristic of the Holstein breed, with a few exceptions of individuals with red- and white-spotted coats. In contrast, the coat color of the Jersey breed is mostly fawn-, mulberry-, or gray-colored (Porter, 1991). Photographs of Holstein and Jersey cows are shown in Figure 1. The bovine melanocortin-1 receptor (MC1R) gene, also known as the extension (E) locus, controls the production of black (eumelanin) and red (pheomelanin) pigments (Klungland et al., 1995). There are three known alleles at this locus: dominant black (E_D, c.296 T > C), wild type (E⁺), and recessive red (e, c.310 G > Deletion). The E^D allele is dominant to the other two, and animals with this allele have a black coat color. Cattle with the E⁺/E⁺ and E⁺/e genotypes have reddish brown to brownish black coloration, while those homozygous for the e allele have a red coat color (Seo et al., 2007). These polymorphisms of the MC1R gene have also been previously utilized for the traceability of Japanese beef (Sasazaki et al., 2007).

Fig. 1.

Photograph of Jersey (left) and Holstein (right) cows. Jersey cows are characterized by a brown coat, while the overwhelming majority of Holstein cattle have a black coat with white spots. Polymorphisms of the MC1R gene are responsible for the variations in bovine coat color.

To provide an effective method for detecting the contamination of Jersey mono-labeled milk with Holstein milk, the distribution of these three alleles at the MC1R locus was investigated in Japanese Holstein and Jersey cow populations. Using a DNA extraction kit for blood samples, it was possible to extract DNA for genetic testing from various packaged milk items. This DNA-based test demonstrated that the contamination of Holstein milk is easily detectable by genotyping of the polymerase chain reaction (PCR) product of the partial MC1R gene.

Materials and Methods

Cattle DNA samples    Hair samples were collected from 204 individual cows (153 Holstein and 51 Jersey) on dairy farms in Shizuoka and Kanagawa Prefectures in Japan. The coat colors of all 153 Holsteins were black with white spots. DNA was extracted from the hair root using ISOHAIR DNA Extraction Kit (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan). As red-haired cattle controls, DNA samples were also collected from three individual Hereford cows.

DNA extraction from commercial milk items    Eight commercially available milk items in paper cartons were purchased and investigated as shown in Table 1. Three were pure Jersey milk, each of which was surveyed five times on different dates. The remaining five milk items did not specify a breed on the label, which were assumed to mostly contain Holstein milk.

Table .1 Product display of surveyed paper carton milk samples.

Item ID	Display of breeds	Sterilization or pasteurization method	Manufacturing company ID 1	Location (Prefecture)	Note
N1	None	130 °C 2 s	A	Hokkaido
N2	None	125 °C 2 s	B	Nagano
N3	None	140 °C 2 s	C	Hokkaido	Shelf-Stable
N4	None	66 °C 30 min	D	Iwate
N5	None	65 °C 30 min	E	Gunma	Organic
J1	Jersey	125 °C 2 s	B	Nagano
J2	Jersey	130 °C 2 s	F	Iwate
J3	Jersey	65 °C 30 min	E	Gunma

1  The same symbol indicates one dairy company.

DNA was extracted directly from 200 µL of milk or from 200 µL chloroform-extracted aqueous layer of milk using QuickGene DNA whole blood kit S (FUJIFILM Wako Pure Chemical Corporation). Chloroform extraction was performed to remove milk fat. An equal volume of chloroform/3-methyl-1-butanol (24:1, v/v) solution was added to 500 µL of milk samples, and the mixture was vortexed for 1 min. The mixture was centrifuged at 5 000 × g for 10 min. Then, 200 µL of the water phase was collected, and DNA was extracted. The detection sensitivity of Holstein contamination was examined with the use of calibration standards prepared by mixing Jersey milk and other milk without a specified breed label in various ratios.

Genotyping of bovine MC1R alleles    Restriction fragment length polymorphism (RFLP) analysis was performed to genotype the bovine MC1R alleles. Partial MC1R gene fragments were amplified by PCR using gene-specific primers previously reported for the genotyping of bovine coat color (Klungland et al., 1995; Sasazaki et al., 2005). A 155-bp fragment was amplified by the F202 (5′-AACCTGCACTCCCCCATGTACTACT-3′) and R356 (5′-ACATTGTCCAGCTGCTGCACCACGG-3′) primer pair (Sasazaki et al., 2005), and a 739-bp fragment was amplified by the E3 (5′-GTGCCTGGAGGTGTCCATC-3′) and E4 (5′-GAAGTTCTTGAAGATGCAGCC-3′) primer pair (Klungland et al., 1995). PCR was performed in a 20-µL mixture, containing 1 µL DNA template, 1X KAPA Taq Extra Buffer, 2.5 mmol/L MgCl₂, 300 µmol/L dNTPs, 500 pmol/L of each primer, and 0.5 U of KAPA Taq Extra DNA Polymerase (Kapa Biosystems, Wilmington, MA, USA). After 5 min of denaturation at 94 °C, PCR comprised 35 cycles of denaturation at 94 °C for 30 s, annealing at 63 °C for 15 s, and extension at 72 °C for 20 s (155 bp)/45 s (739 bp), and a final extension step at 72 °C for 5 min using TaKaRa PCR Thermal Cycler Dice^® (Takara Bio, Inc., Shiga, Japan). For RFLP analysis, approximately 100 ng of the PCR product was incubated with 2 U of restriction endonuclease and CutSmart buffer in a total volume of 20 µL (New England Biolabs, Ipswich, MA, USA). The PCR-amplified fragments generated using the F202 and R356 primer pair (155 bp) were used to distinguish the E^D allele from the E⁺ and e alleles (Sasazaki et al., 2005). The 155-bp PCR products were digested with the AciI restriction enzyme (New England Biolabs), which resulted in cleavage of the E^D allele into 60 bp and 95 bp fragments, with no accompanying change in the amplicon of the E⁺ and e alleles. The amplification products produced with the E3 and E4 primer pair (738 or 739 bp) were used to distinguish the e allele from the E^D and E⁺ alleles (Klungland et al., 1995). The amplicons generated by the E3 and E4 primer pair were digested with 2 U of the BsrFI-v2 restriction enzyme (New England Biolabs), which resulted in cleavage of the E^D and E⁺ alleles into 208-bp and 531-bp fragments, while the e allele was uncleaved and remained at 738 bp.

These digested amplicons were separated by electrophoresis on 2% or 3% agarose gels with Tris-borate-ethylenediaminetetraacetic acid buffer. These gels were stained with ethidium bromide and viewed under ultraviolet light. The three alleles of MC1R included E^D, which was cut by both AciI and BsrFI-v2, E⁺, which was cut by BsrFI-v2 but not AciI, and e, which was not cut by either enzyme.

Results and Discussion

Allelic frequencies of the MC1R gene in dairy cattle in Japan    RFLP analysis (Figure 2) was performed to distinguish the three main alleles of the bovine MC1R locus. The genotypic and allelic frequencies at the MC1R locus of the analyzed Holstein and Jersey cow populations are shown in Table 2. All Holstein samples harbored at least one copy of the E^D allele. The allelic frequencies of the E^D, E⁺, and e alleles in the Holstein samples were 0.876, 0.118, and 0.007, respectively. In the Jersey samples, the frequency of the E⁺ allele was quite high, while that of the e allele was very low (0.98 and 0.02, respectively). These allelic frequencies were nearly in agreement with a report of Jersey cattle in Italy (Russo et al., 2007). The E^D allele was not detected in any samples from Jersey cows. Detection of the E^D allele clearly indicates the presence of Holstein-derived ingredients in dairy products.

Fig. 2.

PCR-RFLP genotyping results of bovine MC1R alleles. Digested PCR products were separated by electrophoresis on 2% agarose gels and stained with 0.5 µg/mL of ethidium bromide. The letter M in each panel indicates a 100-bp DNA Ladder molecular weight marker (New England Biolabs). The genotype is shown above each lane. (A) Genotyping to distinguish E^D from the other alleles (E⁺ and e). (B) Genotyping to distinguish the e allele from the other alleles (E^D and E⁺). The bands marked with * in are nonspecific amplicons that had no effect on genotyping.

Table 2. Genotypic and allelic frequencies of the MC1R (coat color extension locus) for cattle breeds.

		Genotype frequencies						Allele frequencies
Breed	N	ED/ED	ED/E+	ED/e	E+/E+	E+/e	e/e	ED	E+	e
Holstein	153	115	36	2	0	0	0	0.876	0.118	0.007
Jersey	51	0	0	0	49	2	0	0.000	0.980	0.020
Hereford	3	0	0	0	0	0	3	0.000	0.000	1.000

In the Holstein samples in this study, the frequency of the E^D allele was 0.876, similar to reports on Holstein in Italy (0.89) (Russo et al., 2007) and in the USA (0.94) (Lawlor et al., 2014). Of note, in Italy and the USA, the e allele was the second most common (0.11 in Italy and 0.05 in USA) (Russo et al., 2007; Lawlor et al., 2014), whereas the frequency of the e allele was only 0.007 in this study (Table 2). Since the cattle hair samples were collected in a limited area, it cannot be said that it is a randomized specimen that can represent the whole of Japan. The 153 Holstein individuals analyzed in this study were not sufficient to reveal the gene frequency of the Holstein population throughout Japan but may be adequate to determine possible trends. Sasazaki et al. (2005) also reported the absence of the e allele in 146 Holstein cows in Japan, although the study did not distinguish between the E^D and E⁺ alleles. Nonetheless, these results suggested that the allelic frequency of e was lower in Holstein populations in Japan in comparison to European and American populations. Furthermore, the E⁺ allele, which has a reported frequency of < 0.01 in Holstein cows in Italy and the USA (Russo et al., 2007; Lawlor et al., 2014), was 0.118 in Japanese Holstein in this study. The relatively high frequency of the E⁺ allele may be due to the history of Holstein breeding in Japan. The Holstein breed was first introduced to Japan from the USA and Europe in the late 19^th century; subsequently, the Holstein breed was crossed with cattle native to Japan. The Holstein population in Japan was established by a continual backcrossing of Holstein bulls to crossbred cows (Tsuji et al., 2004), which impacted the genetic composition of Holsteins in Japan. In fact, approximately 18% of Holstein cattle in Japan have a unique mitochondrial DNA haplotype, which is not present in Holsteins in the USA but is abundant in Japanese-Black breed (Tsuji et al., 2004). As the E⁺ allele is most common both in Japanese-Black (0.514) and Japanese-Brown (0.962) breeds, which are called Wagyu and are unique beef cattle in Japan (Sasazaki et al., 2005), the presence of the E⁺ allele in Holsteins in Japan might have originated from native ancestral cattle populations.

Traceability of Jersey milk in packaged milk items in Japan    Because 1 mL of raw cow's milk contains an average of approximately 1 × 10⁵ somatic cells (Olechnowicz and Jaskowski, 2012), milk contains large amounts of cow DNA (Lipkin et al., 1993, Russo et al., 2007). Here, a silica membrane-based DNA purification kit designed for small-scale blood sampling was tested to extract DNA from commercial packaged milk items. We were able to extract DNA from commercial packaged milk items simply by replacing blood with milk without changing the kit protocol. On a technical note, insufficient mixing of lysates sometimes clogged the column with insoluble matter; however, vortexing at high speed for 1 min before loading the lysate onto the column prevented blockage effectively. When DNA was extracted from the aqueous fraction of chloroform-treated milk, no column blockage occurred, and the quality of DNA was similar to that of the direct use of milk.

The results of RFLP analysis for detecting the E^D allele in eight commercially available milk items are shown in Figure 3. PCR amplification was sufficient for all items, and genotyping was possible. These milk-based products were packaged after pasteurization at 65 °C for 30 min or sterilization at 125 °C–140 °C for 2 s (Table 1). These results indicated no significant damage to DNA during the milk sterilization process.

Fig. 3.

Detection of E^D and non-E^D alleles of the MC1R gene in milk items. N1–N5 are milk items without a breed label, and J1 to J3 were labeled as pure Jersey milk items (Table 1). N1, N2, J1, and J2 are ultra-high temperature processed (UHT) milk, and N3 is long-life (LL) milk. N4, N5, and J3 are pasteurized milk items.

DNA fragments of the E^D allele cleaved with the AciI enzyme were detected in all five milk items without a breed label, although the ratios of the E^D and other alleles varied (Figure 3). Detection of the E^D allele indicates the presence of Holstein milk because this allele is not present in non-black coated cows. In contrast, the E^D allele was not detected in three items containing milk from Jersey cows (Figure 3). Hence, it was possible to distinguish dairy products containing pure milk from Jersey cows and from products with no breed label containing milk from Holstein cows. Of the dairy products containing only milk from Jersey cows, the E^D allele was not detected in five trials conducted on different production dates. These results confirmed that dairy companies in Japan strictly adhere to the use of pure Jersey milk to prevent contamination with Holstein milk. In three of the five milk items without a breed label, E^D and non-E^D (E⁺ and e) DNA fragments were detected (Figure 2). The detected non-E^D allele in breed-unspecified milk items might have originated from not only Holstein cows carrying a non-E^D allele but also Jersey cows, as some dairy farms raise both breeds and milk them together.

The detection sensitivity of contaminated Holstein milk was examined using dairy products containing mixtures of Jersey milk (J1 of Figure 3) and breed-unspecified milk (N1 of Figure 3) (Figure 4). The E^D allele was detected in dairy products containing ≥ 10% of breed-unspecified milk mixed with Jersey milk.

Fig. 4.

Detection of E^D and non-E^D alleles in mixed samples of Jersey milk and unlabeled milk items. DNA was extracted from mixed milk in various proportions of N1 and J1, as shown in Figure 3. PCR was performed with the primer pair F202 and R356. PCR products were digested with the restriction enzyme AciI and then separated on a 3% agarose gel. The E^D allele was detected in samples containing ≥ 10% of unlabeled milk.

Limitations of the study.    In the present study, we used a conventional method to identify RFLPs in PCR endpoint products using agarose gel electrophoresis. Measurements at the PCR endpoint could not be quantitated due to the plateau effect. In addition, our conventional method requires more time and is less sensitive than quantitative real-time PCR, which is currently widely used in food inspection. Introducing a real-time PCR system using an allele-specific fluorescent reporter probe for genotyping the bovine MC1R gene will shorten the test time and improve sensitivity (Royo et al., 2008; Navarro et al., 2015). Milk in paper cartons is used not only for drinking but also as an ingredient for making bread and sweets. Unlabeled milk may be used in confectioneries to make cakes and puddings labeled with Jersey milk. Hence, it is necessary to further improve the genetic testing method to distinguish Jersey and Holstein milk products both in milk and in processed foods, such as confectioneries.

Acknowledgements    The authors are grateful to Takahiro Ito for his valuable technical assistance on DNA analysis. The authors would like to thank Enago (www.enago.jp) for the English language review.

Conflicts of interest    The authors have no conflicts of interest to declare.

List of Abbreviations

PCR

polymerase chain reaction

RFLP

restriction fragment length polymorphism

References

•  Dalvit,  C.,  De Marchi,  M., and  Cassandro  M. (2007). Genetic traceability of livestock products: A review. Meat Sci., 77, 437-449.
•  Kijas,  J.M.,  Wales,  S.R.,  Tornsten,  A.,  Chardon,  P.,  Moller,  M., and  Andersson,  L. (1998). Melanocortin receptor 1(MC1R) mutations and coat color in pigs. Genetics., 150, 1177-1185.
•  Klungland,  H.,  Våge,  D.I.,  Gomez-Raya,  L.,  Adalsteinsson,  S., and  Lien,  E. (1995). The role of melanocyte-stimulating hormone (MSH) receptor in bovine coat color determination. Mamm. Genome, 6, 636-639.
•  Lawlor,  T.J.,  VanRaden,  P.M.,  Null,  D.,  Levisee,  J., and  Dorhorst,  B. (2014). Using haplotypes to unravel the inheritance of Holstein coat color. Proceedings of the 10th World Congress on Genetics Applied to Livestock Production. Vancouver, Canada, Volume Species Breeding: Dairy cattle, p. 289.
•  Lipkin,  E.,  Shalom,  A.,  Khatib,  H.,  Soller,  M., and  Friedmann,  A. (1993). Milk as a source of deoxyribonucleic acid and as a substrate for the polymerase chain reaction. J. Dairy Sci., 76, 2025-2032.
•  Maudet  C. and  Taberlet  P. (2002). Holstein's milk detection in cheeses inferred from melanocortin receptor 1 (MC1R) gene polymorphism. J. Dairy Sci., 85, 707-715.
•  Navarro,  E.,  Serrano-Heras,  G.,  Castaño,  M.J., and  Solera,  J. (2015). Real-time PCR detection chemistry. Clin. Chim. Acta., 439, 231-250.
•  Olechnowicz,  J., and  Jaskowski,  J.M. (2012). Somatic cells count in cow's bulk tank milk. J. Vet. Med. Sci., 74, 681-686.
•  Porter  V. (1991). “Cattle - a Handbook to the Breeds of the World.” Christopher Helm, London, UK.
•  Rouzaud,  F.,  Martin,  J.,  Gallet,  P.F.,  Delourme,  D.,  Goulemot-Leger,  V.,  Amigues,  Y.,  Ménissier,  F.,  Levéziel,  H.,  Julien,  R., and  Oulmouden,  A. (2000). A first genotyping assay of French cattle breeds based on a new allele of the extension gene encoding the melanocortin-1 receptor (MC1R). Genet. Sel. Evol., 32, 511-520.
•  Royo,  L.J.,  Fernández,  I.,  Azor,  P.J.,  Alvarez,  I.,  Pérez-Pardal,  L., and  Goyache,  F. (2008). Technical note: a novel method for routine genotyping of horse coat color gene polymorphisms. J. Anim. Sci., 86, 1291-1295.
•  Russo,  V.,  Fontanesi,  L.,  Scotti,  E.,  Tazzoli,  M.,  Dall'Olio,  S., and  Davoli,  R. (2007). Analysis of melanocortin 1 receptor (MC1R) gene polymorphisms in some cattle breeds: their usefulness and application for breed traceability and authentication of Parmigiano Reggiano cheese. Ital. J. Anim. Sci., 6, 257-272.
•  Sasazaki,  S.,  Usui,  M.,  Mannen,  H.,  Hiura,  C., and  Tsuji,  S. (2005). Allele frequencies of the extension locus encoding the melanocortin-1 receptor in Japanese and Korean cattle. Anim. Sci. J., 76, 129-132.
•  Sasazaki,  S.,  Mutoh,  H.,  Tsurifune,  K., and  Mannen,  H. (2007). Development of DNA markers for discrimination between domestic and imported beef. Meat Sci., 77, 161-166.
•  Sasazaki,  S.,  Hosokawa,  D.,  Ishihara,  R.,  Aihara,  H.,  Oyama,  K., and  Mannen,  H. (2011). Development of discrimination markers between Japanese domestic and imported beef. Anim. Sci. J., 82, 67-72.
•  Seo,  K.S.,  Mohanty,  T.R.,  Choi,  T., and  Hwang,  I.H. (2007). Biology of epidermal and hair pigmentation in cattle: a mini review. Vet. Dermatol., 18, 392-400.
•  Tsuji,  S.,  Mannen,  H.,  Mukai,  F.,  Shojo,  M.,  Oyama,  K.,  Kojima,  T.,  Kano,  C.,  Kinoshita,  Y., and  Yamaguchi,  E. (2004). Trace of native cattle in Japanese Holstein assessed by mitochondrial DNA sequence polymorphism. J. Dairy Sci., 87, 3071-3075.
•  Olechnowicz,  J., and  Jaskowski,  J.M. (2012). Somatic cells count in cow's bulk tank milk. J. Vet. Med. Sci., 74, 681-686.
•  Porter  V. (1991). “Cattle - a Handbook to the Breeds of the World.” Christopher Helm, London, UK.
•  Rouzaud,  F.,  Martin,  J.,  Gallet,  P.F.,  Delourme,  D.,  Goulemot-Leger,  V.,  Amigues,  Y.,  Ménissier,  F.,  Levéziel,  H.,  Julien,  R., and  Oulmouden,  A. (2000). A first genotyping assay of French cattle breeds based on a new allele of the extension gene encoding the melanocortin-1 receptor (MC1R). Genet. Sel. Evol., 32, 511-520.
•  Royo,  L.J.,  Fernández,  I.,  Azor,  P.J.,  Alvarez,  I.,  Pérez-Pardal,  L., and  Goyache,  F. (2008). Technical note: a novel method for routine genotyping of horse coat color gene polymorphisms. J. Anim. Sci., 86, 1291-1295.
•  Russo,  V.,  Fontanesi,  L.,  Scotti,  E.,  Tazzoli,  M.,  Dall'Olio,  S., and  Davoli,  R. (2007). Analysis of melanocortin 1 receptor (MC1R) gene polymorphisms in some cattle breeds: their usefulness and application for breed traceability and authentication of Parmigiano Reggiano cheese. Ital. J. Anim. Sci., 6, 257-272.
•  Sasazaki,  S.,  Usui,  M.,  Mannen,  H.,  Hiura,  C., and  Tsuji,  S. (2005). Allele frequencies of the extension locus encoding the melanocortin-1 receptor in Japanese and Korean cattle. Anim. Sci. J., 76, 129-132.
•  Sasazaki,  S.,  Mutoh,  H.,  Tsurifune,  K., and  Mannen,  H. (2007). Development of DNA markers for discrimination between domestic and imported beef. Meat Sci., 77, 161-166.
•  Sasazaki,  S.,  Hosokawa,  D.,  Ishihara,  R.,  Aihara,  H.,  Oyama,  K., and  Mannen,  H. (2011). Development of discrimination markers between Japanese domestic and imported beef. Anim. Sci. J., 82, 67-72.
•  Seo,  K.S.,  Mohanty,  T.R.,  Choi,  T., and  Hwang,  I.H. (2007). Biology of epidermal and hair pigmentation in cattle: a mini review. Vet. Dermatol., 18, 392-400.
•  Tsuji,  S.,  Mannen,  H.,  Mukai,  F.,  Shojo,  M.,  Oyama,  K.,  Kojima,  T.,  Kano,  C.,  Kinoshita,  Y., and  Yamaguchi,  E. (2004). Trace of native cattle in Japanese Holstein assessed by mitochondrial DNA sequence polymorphism. J. Dairy Sci., 87, 3071-3075.
• i) https: //www.id.nlbc.go.jp/data/toukei.html (Jul. 10, 2020)